Subject: Re[2]: [linux-audio-dev] Performance and Elegance? (Was: High Cost of
From: Rick Burnett (destinytech_AT_spacey.net)
Date: Thu May 17 2001 - 03:07:20 EEST
I wanted to point out something I discovered on Sun systems pertaining
to memory. At the company I work for we bought 5 new Ultra 80s, which
I thought were pretty good not counting the new blade systems. We
also maxed out the memory. I was looking at the memory and discovered
it was 50ns!! I thought this cannot be! Then I discovered that
instead of using faster RAM, the memory bus is 520 bytes wide ( which
includes some parity ). Now I have done no research on the PC
architecture, anyone have comments on this?
Rick
Wednesday, May 16, 2001, you wrote:
JT> Rob Melby wrote:
>>
>> main memory will always be slower than registers.
JT> Yes, I expect so. I was only pointing out that there is nothing
JT> about basic memory technology (the transisitors themselves, and the
JT> memory circuits made from them) that is so much slower (like, 10x
JT> slower) than CPU speeds. So it is possible to make RAM that is
JT> significantly faster than PC133 memory that is now common.
>> IPC is a bad way to go compared to in-process methods for a number of
>> reasons.
JT> If we have to pick betwen the two, then I surely agree. For performance
JT> reasons, anyway. For other reasons, IPC may be better, although the
JT> absolute need for low latency may make that a dead issue. Or maybe not ...
JT> I don't think we've touched on all the important factors regarding
JT> this yet.
JT> - Jay Ts
JT> jayts_AT_bigfoot.com
This archive was generated by hypermail 2b28 : Thu May 17 2001 - 03:19:41 EEST